Modern computer systems typically consist of a number of different computers and computer peripherals interconnected by a network. Computer networking is a rapidly changing technology, making it challenging for vendors to keep the technology of computer related products current with computer networking technology. Products need to be updated frequently for at least two reasons, namely, to adapt to additional networking topologies, and to provide new features demanded by the marketplace.
To meet the demand for rapidly changing computer products, field configurable embedded computer systems were introduced. A field configurable embedded computer system consists of a small computer system included in a consumer or business product. A field configurable embedded computer system typically contains a central processing unit, a persistent memory area for storing programs and persistent variables that need to be preserved when the product is powered down, a volatile memory area that is operational while the unit containing the field configurable embedded computer system is powered on, but is erased when the unit is powered off, and a network communications hardware means, including the associated software drivers that enable the device to connect to a computer network.
Field configurable embedded computer systems are used to build a family of products. Each product in the family uses the same field configurable computer system hardware. The family members differ from each other in the feature functions they provide. The feature functions for a particular family member are implemented to computer software. Thus by changing the software, a product can be changed from one family member to another.
A field configurable embedded computer systems is said to be configured as a particular family member when the appropriate software for that family member has been installed in the field configurable embedded computer system. The computer software is installed on a particular field configurable embedded computer system by transferring a copy of the software from a master computer to that particular field embedded computer system. The process of transferring computer software from a master computer to a particular field configurable embedded computer systems is known as downloading.
Downloading can take place over a computer network such as the internet. Thus the master computer can be at the manufacturer""s factory, and the field configurable embedded computer system can be simultaneously installed at a remote customer site. More commonly, the new download code set is transferred to the customer via the internet or other means, and a customer computer is used as the master computer. If the master computer and the field configurable embedded computer system are connected to the same network such as the internet or a company private network, then a download can take place from the master computer to the unit while it remains installed in the customer site as long as the download code is operational in the device. This is very convenient because the customer does not have to return the product to the factory for an upgrade or even shut it off or disconnect it from the network. In extreme cases, downloading is used to totally configure a field configure embedded computer system from one family member to a completely different family member, effectively swapping products.
Field configurable embedded computer systems operate in two basic modes. The first mode is the product""s normal operating mode where it functions as a useful product. The second mode is a download mode where the product in the field is poised to receive new computer code from a master computer at the factory by means of a download over the network. After the download is complete, the product is then returned to its normal operating mode.
Unfortunately, in traditional field configurable embedded computer systems the download program code section of the persistent memory area tends to be large because multiple networking environments utilized by various family members of the product line need to be supported. This results in at least three problems. The first problem is that although the product family generally supports a large number of different networking environments, a particular family member only needs to support a single networking environment. The remaining network drivers in the download network driver sections are ineffective, wasting code space in the program download code section that could be better used to provide additional feature functions. The second problem is that the download network driver section is largely redundant with the feature network driver section, because both provide a means to communicate with supported computer networks. This inefficient use of the persistent memory area leaves less room in the persistent memory area for feature functions. A third problem with existing field configurable embedded computer systems is that since only the feature code section can be reconfigured by a download, the download program code section itself cannot be upgraded.
Updating feature code from one computer from another computer is well known. U.S. Pat. No. 5,142,680 to Ottman (the ""680 patent) teaches how to upgrade an operating system from one computer to another. U.S. Pat. No. 5,261,055 to Moran (the ""055 patent) teaches of a generalized scheme to update computer code using a separate, dedicated processor for the download functions. U.S. Pat. No. 5,404,450 to Szczepanak (the ""450 patent) teaches how to load new networking protocols on computers on a network. U.S. Pat. Nos. 5,261,114 and 5,408,624 to Raasch (the Raasch patents) teach of a method of updating code using the ISA bus instead of a network. U.S. Pat. No. 5,239,621 to Brown (the ""621 patent) teaches of a variation of a field configurable embedded computer system where the feature function are fonts and macros for a printer. U.S. Pat. No. 5,402,528 to Christopher (the ""528 patent) teaches how to update specific printer functions by partitioning memory into two sections and only the second can be upgraded.
Although the above cited patents teach various methods of updating one computer from another computer, none teach any specific methodology to update the downloading mechanism itself. The ""680 is directed toward large computers rather than embedded computers, and the core downloading means taught by the ""680 patent must be loaded on to the transferee from a source other than the transferror computer, providing no way to update the download means itself. The ""621 patent uses the existing datastream to perform the download, thus the downloading mechanism itself cannot be altered in the field. The ""055 patent uses a separate, dedicated and permanent downloading means. The ""450 patent does not provide a way to change the downloading protocol itself; in fact the downloading mechanism is protected from change in the ""450 teachings. The Raasch patents to not provide a means for updating the download code itself; furthermore the Raasch teachings are not readily applicable to a networking environment and are directed devices contained within a single computer. The ""528 patent does not teach how to update the download mechanism itself, and it provides no means to upgrade the first memory section.
The ESI 2830 Pocket Printer server (the ""2830) is similar to the ""528 patent in that it provides two distinct memory sections: one for downloading code and another for features functions for the print server. The ""2830 is an improvement over the ""528 patent in that unlike the ""528 patent, both memory sections can be updated by a download. However, the ""2830 scheme requires dedicated networking code in its downloading means separate from and redundant with large amounts of networking code in the feature functions of the second memory section, thus severely limiting the features of the product.
What is needed is a field configurable embedded computer system that both eliminates redundant network driver code and allows new network driver code, including the download code, to be downloaded.
The present invention replaces both the download network driver section and the feature network driver section with a single combined network driver section, thus eliminating redundant network drivers. The present invention allows the download program code itself to be updated during a download, by dividing the download code set into different sections and using both persistent memory and volatile memory in a novel way.
Accordingly, it is an object of the present invention to provide a single set of network driver code for both downloading purposes and implementation of feature functions. It is a further object of the present invention to provide a means to download new download code, including new network drivers, to a field configurable embedded computer system installed at a customer site in the field. It is a further object of the present invention to minimize the amount of time critical code is executing out of volatile memory thus minimizing the likelihood of a catastrophic failure during download that would require factory repair.